Claude Opus 5 (Adaptive Reasoning, High Effort): benchmark results

Provider: Anthropic. Released 2026-07-24. Access: API.

Unified ELO 1720 ± 1, rank #63 of 1761 rated models, from 20 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA Omniscience - Science, Engineering & Mathematics56.21Accuracy (%)98.6
Artificial Analysis Intelligence Index52.03Intelligence Index98.4
AA Humanity's Last Exam52.83Accuracy (%)98.3
AA Omniscience - Business50.7Accuracy (%)98.2
AA GPQA Diamond93.74Accuracy (%)98
AA Omniscience33.72Score97.5
AA Omniscience - Health51.76Accuracy (%)97.3
AA-Omniscience Accuracy58.88Accuracy (%)97.3
AA CritPt28.29Accuracy (%)96.9
AA Omniscience - Humanities & Social Sciences56Accuracy (%)96.5
AA Omniscience - Software Engineering (SWE)83.9Accuracy (%)96.5
AA Omniscience - Law55.2Accuracy (%)94.7

Interactive version: theaggregate.ai/model?slug=claude-opus-5-adaptive-reasoning-high-effort · How It Works · Data refreshed daily, snapshot 2026-09-05.